As artificial intelligence moves from creating advertising to making media decisions, marketers face a new question: how much control are they prepared to hand over to machines, and who takes responsibility when automated systems make the wrong call?
AI is already embedded in digital advertising, helping marketers target audiences, bid for inventory and optimise campaigns. The next phase involves agentic AI: systems that can connect research, planning, execution and measurement, making decisions across a campaign with less human intervention.
The shift is attracting attention from global marketing leaders. A Boston Consulting Group survey of 300 chief marketing officers, published in June 2026, found that 96% said AI was driving end-to-end transformation of their marketing functions. Yet only about a third had completed the work needed to achieve that transformation. Just 8% reported running campaigns in which multiple AI agents operated autonomously.
The figures suggest that while enthusiasm is widespread, fully autonomous marketing remains at an early stage.
When automation meets brand safety
The challenge is not simply whether AI can allocate money efficiently. It is whether it can make decisions that protect a brand’s reputation while meeting performance targets.
Traditional brand-safety systems help advertisers avoid placements alongside harmful, offensive or otherwise unsuitable content. But as AI agents assume broader responsibilities, the decision-making chain becomes more complicated. A brand sets the objectives, an agency configures the campaign, a platform delivers the advertising, and an AI system may determine where spending is directed.
If an agent shifts a budget towards a high-performing but reputationally risky placement, who is accountable: the brand, its agency, the technology provider or the platform?
The question is becoming more immediate as advertising enters conversational AI environments. OpenAI launched its ChatGPT advertising pilot in the United States in February 2026, initially targeting users of its Free and Go plans. The rollout attracted major agency groups, including Omnicom, WPP and Dentsu. According to Adweek, Omnicom Media said more than 30 of its clients had secured placements in the pilot.
On October 5, OpenAI announced expanded advertising measurement and new brand-suitability initiatives, including evaluation pilots with DoubleVerify and Integral Ad Science (IAS). The independent verification companies will help advertisers assess how ChatGPT applies its advertising safeguards, without accessing private user conversations.
The move highlights a central issue for the industry: advertisers want independent evidence that platforms are applying their safeguards consistently, rather than relying exclusively on assurances from the companies selling the media.
However, verifying where an advertisement appears is only part of the challenge. As agents gain more autonomy, advertisers will also need to understand why budgets were moved, which rules governed those decisions and whether performance objectives compromised brand suitability.
What it means for Gulf marketers
The debate is particularly relevant to the Middle East, where international brands must balance global brand-safety standards with local cultural sensitivities and market-specific requirements.
For Gulf chief marketing officers and agency leaders, the immediate question is whether any organisations are already allowing AI agents to reallocate media budgets without human approval, and what safeguards govern those decisions.
Practical protections could include spending limits, approved placement lists, human sign-off for significant budget changes and auditable records of automated decisions.
The industry must also distinguish between existing programmatic advertising, which automates bidding under predefined rules, and agentic systems that can make broader decisions across workflows.
Ultimately, the question is not whether AI can buy advertising more efficiently. It is whether brands can retain meaningful control over where their money goes, how their reputations are protected and who answers when an autonomous system gets it wrong.
